12.31.21 report

Hill numbers diversity

This is diversity measured by hill number indices where increasing hill number (Q) corresponds with more focus on the largest clonotypes. For example, Q=0 is just the number of clonotypes, Q=1 is shannon index, Q=2 is the inverse simpson index, Q=\(\infty\) is 1/(proportion of the largest clone).

In pretreatment, Lynch generally has the highest clonality (lowest diversity) through all Q values although when putting more weight on the very largest clonotypes, NLR is similarly clonal indicating that Lynch actually doesn’t have higher clonality solely by a few extremely large clonotypes
In posttreatment, NLR generally has the highest clonality in higher Q values while Lynch is only slightly more clonal than NR, although the pattern is reversed at Q=1 (shannon-wiener entropy)
Healthy always has the highest diversity

Misc diversity stuff

This is measuring the diversity of the PBMC TIL matching repertoire

Pretreatment seems to follow NLR more clonal than L more clonal than NR, but posttreatment L and NR seem similar
NLR seems more clonal indicating that the TIL matching clones in the peripheral blood are more expanded both before and after treatment
Turnover?

This is the clonality of the TILs themselves (all pretreatment)

NR is most clonal and L is least with NLR somewhere inbetween
Clonality delta NR?

TM prop

This is change over time in the proportion of peripheral blood cells that have TIL matching TCRs

NLR has a greater tendency towards an increase in TIL matching
What does this mean when the TILs this is matching to are the pretreatment TILs?
Expansion in novel or preexisting TM clones?

Rarefaction

These are just the rarefaction curves from last week:

Lynch has much more large clonotypes in pretreatment and more posttreatment (but not as much more). NLR has the second largest general clone sizes followed by NR and then H.
Ultimately, most of this is kind of similar to looking at an intermediate to large hill number except more easily interpretable

This is specific for TIL clone sizes and as with the hill numbers, NR are much more clonal

VJ gene stuff

All of these are pretreatment - the first 4 lynch, the next 4 NLR, the last 4 some of the nonresponders

Not sure how we want to represent this since there is a lot of variation among the patients. There are expanded L patients as well as one for NR that’s pretty expanded, kinda variable.

## # A tibble: 15 x 7
##    CDR3.aa             V.name   J.name  Clones Proportion expTil cluster_fine
##    <chr>               <chr>    <chr>    <int>      <dbl>  <dbl> <chr>       
##  1 CASSVAPSGDGGGTGELFF TRBV9    TRBJ2-2   1042     0.186      34 CD8 TEMRA   
##  2 CASSTGAGDTYEQYF     TRBV9    TRBJ2-7    503     0.0898    217 CD8 TEMRA   
##  3 CASSDYNEQFF         TRBV11-2 TRBJ2-1    486     0.0868     47 CD8 TEMRA   
##  4 CASSSYNEQFF         TRBV11-2 TRBJ2-1    192     0.0343     60 CD8 TEMRA   
##  5 CASSPITGTGHYGYTF    TRBV6-5  TRBJ1-2    161     0.0288     95 CD8 TEMRA   
##  6 CSVRANTEAFF         TRBV29-1 TRBJ1-1    154     0.0275    168 CD8 TEMRA   
##  7 CASSLVTGTGSYGYTF    TRBV6-5  TRBJ1-2    153     0.0273     35 CD8 TEMRA   
##  8 CASSERPDYGYTF       TRBV2    TRBJ1-2    111     0.0198     61 CD8 TEMRA   
##  9 CASSELGTGYDEQFF     TRBV2    TRBJ2-1    110     0.0196     46 CD8 TEMRA   
## 10 CASSVEAGENTDTQYF    TRBV9    TRBJ2-3    103     0.0184     19 CD8 TEMRA   
## 11 CASSYAASYEQYF       TRBV6-5  TRBJ2-7     92     0.0164      9 CD8 TEMRA   
## 12 CASSFGDEQYF         TRBV11-3 TRBJ2-7     77     0.0138    113 CD8 TEMRA   
## 13 CSVFRGDRYGYTF       TRBV20-1 TRBJ1-2     69     0.0123      2 CD8 TEMRA   
## 14 CAWSVQVEAFF         TRBV30   TRBJ1-1     60     0.0107     21 CD8 TEMRA   
## 15 CASSQYNEQFF         TRBV11-2 TRBJ2-1     59     0.0105     63 CD8 TEMRA

Sidenote, the super expanded chord for Lynch-like pretreatment (TRBV9.TRBJ2-2) is all one clonotype from 14-1 which is almost 20% of all cells.

The second most expanded chord though is also from 14-1 (TRBV11-2.TRBJ2-1), but it’s mostly three different expanded clonotypes which each differ by only one amino acid (in the same position) and are all expanded in TILs as well.

The raw VJ gene diversity is not super different and is very skewed especially pretreatment.

CD4/CD8 specific

Small sidenote: all the points in this section that just have timepoint 1 and 5 - timepoint 5 is a combination of all posttreatment timepoints (3 and 5) that I just labelled 5 for code expediency

Kinda noisy and not generally approaching significance, but Lynch is generally intermediate with NLR as more clonal for CD8 and NR more clonal for CD4. Not sure how much we can read from this pattern since this masks a lot of the more granular cell types.

A bit stochastic, but even with the cell count correction, Lynch like has CD8 TEMRA clonality intermediate between NR and the more clonal NLR. Soooo, Lynch clonality is driven by TEMRA being more clonal and Lynch having more TEMRA, not by Lynch-TEMRA specifically being more clonal.

Hard to pick patterns from T cell one - if anything, NLR might be the most distinct with a lower CD8 pre/posttreatment.

From overall, we can much more clearly see the dominance of CD8 in Lynch-like, but it also makes CD4 and other seem somewhat more even.

NR consistently has lower proportion tumor matching at all timepoints and groups. Lynch has higher CD8 especially pretreatment, less different/interpretable for CD4s.

Density plots

These are all density plot for all cell types pretreatment. Unlike the nebulosa plots, these are calculated for each patient\(x\)timepoint combination and then averaged.

Lynch-like

NLR

Nonresponder

Also worth noting is that I haven’t normalized the color scaling for these, so the highest cell density for nonresponder is about 0.009 while the highest for NLR and lynch are around 0.013, so the graphs are not currently comparable.